The cohort included 20 cases of newly diagnosed LPL for which MYD88 (exon 5) was confirmed via ASO-PCR, and sequencing of CXCR4 C-terminal regulatory cytosolic domain (exon 2) was performed by Sanger sequencing. The median age of the cohort was 69 years (IQR: 63–75) with a 70% male predominance (n = 14).

Serum protein electrophoresis and immune-fixation were performed in 19 cases. The criteria for WM were met in 90% (17 or 19), with a median baseline IgM paraprotein level of 27.5 g/L (IQR: 11.25–44.75), and kappa light chain restriction in 82% (n=14). The complication of autoimmune haemolytic anaemia was biochemically confirmed in two of the eight patients tested. The remaining two cases that did not meet the criteria for WM showed an isolated free kappa light chain in one, while the second showed no evidence of a paraprotein.

Allele-specific oligonucleotide polymerase chain reaction confirmed the MYD88^L265P point mutation in 80% of cases (16 or 20). CXCR4 variants were detected in 45% (n = 9) and were consistently present in conjunction with a MYD88 variant. There were nearly equal numbers of frameshift and nonsense CXCR4 variants, with Table 1 providing a detailed breakdown of each pathogenic variant. A diffuse pattern of tumour infiltration was observed in all histological specimens harbouring CXCR4 nonsense mutations (n = 4), whereas this pattern was present in only two-thirds of cases with CXCR4 frameshift mutations (n = 3). The sample size was too small to reliably assess correlations among mutation status, mutation type and clinical outcomes.

The pertinent demographic, laboratory and survival data for the different mutation subcategories are summarised in Table 2. The majority of cases were MYD88 and CXCR4 mutated (MYD88^MUT CXCR4^MUT) (45%), with around one-third MYD88^MUT CXCR4 wild-type^(WT) and the minority MYD88^WT CXCR4^WT (20%). The lowest peripheral blood counts were evident in the MYD88^MUT CXCR4^MUT group, which also showed the highest male predominance. This subgroup was characterised by a predominantly diffuse pattern of infiltration in nearly 80% of the histological specimens. Serum paraprotein levels were higher in the MYD88^MUT group as compared with the MYD88^WT (p = 0.26), and lower in those who were alive at the time of the study as compared to those who demised; median 21 g/L (IQR: 3–28) versus 39.5 g/L (IQR: 26.75 to 46.25) (p = 0.66). Beta-2 microglobulin (B2M) testing was not routinely performed at diagnosis (it was available for six patients). Applying the simplified risk stratification score proposed by Zanwar et al.,²⁶ high and intermediate-risk scores were predominant among the MYD88^MUT groups, with the MYD88^MUT CXCR4^MUT subgroup showing the highest proportion of high-risk scores.

The crude median OS for the entire cohort was 11.5 months (IQR: 1.9–20.8). Among patients who survived, the median OS was 15 months (IQR: 13–33.9), compared with 3.1 months (IQR: 1.1–13.3) among those who died (p = 0.06). The terminal blood results from patients who died showed a range of abnormalities, including severe pancytopenia or bicytopenia, coagulopathy, elevated markers of infection and renal and/or liver dysfunction. Survivors tended to be younger, with a median age of 63 years (IQR: 55–75) compared with 70 years (IQR: 68–73) among non-survivors, but this difference was not statistically significant (p = 0.52). These findings suggest a trend towards improved survival in younger patients, although the sample size limited the power to detect statistically significant differences. A total of five patients were lost to follow-up at a median of 13.5 months (IQR: 6.5–18.3). This group had a median age of 72 years (IQR: 62–82) and was predominantly male (n = 4, 80%).

The United States Surveillance, Epidemiology, and End Results (SEER) Programme, which has been conducted over 20 years, reported that WM accounted for 1.9% of all NHL cases.²⁸ These statistics are unknown in the South African setting due to both a lack of data and the likely under-recognition of this entity. The SEER data further showed that WM was a disease that primarily affects those of advancing age, with a median age at diagnosis of 73 years and a male predominance of 60%.²⁸ The age of our South African cohort was comparable to SEER, although it included a higher proportion of male patients. The basis for the observed male predominance in LPL or WM remains unclear. Still, it may be attributable to intrinsic sex-specific differences in the immune response, hormone levels, occupational and environmental exposures, or genetic susceptibility.²⁹

The principal genetic alteration in LPL or WM is a somatic mutation in the MYD88 gene, which is almost exclusively the MYD88^L625P variant.¹ Our study revealed a slightly lower proportion of MYD88^L625P mutations as compared to the international literature, which may be attributed to samples with haemodilution or low tumour infiltration, leading to false negative results, or cases misclassified as LPL or WM.²⁷ In MYD88^MUT LPL or WM, secondary co-operating genetic events are necessary for the progression to overt clinical disease. Among the most common are the mutually exclusive recurrent cytogenetic abnormality, del(6q), which is reported in around 40–50% of cases³⁰, and sub-clonal C-terminal CXCR4 mutations, reported in 30–40% of cases.^19,20 While data on the cytogenetic profile of our cohort were limited, the proportion of CXCR4 mutations was slightly higher than reported in the literature.^19,20

The MYD88 and CXCR4 variant status is linked with differences in clinical presentation, therapeutic response and disease outcomes.^19,31 MYD88^WT occurs in older patient groups and is more commonly accompanied by lymphadenopathy. While it is associated with a lower burden of bone marrow disease and serum IgM levels, it is linked to higher levels of poor prognostic markers (such as B2M), inferior OS, increased risk of death and inferior response to ibrutinib monotherapy (no major responses [MR] and shorter progression-free survival [PFS]).^19,32 In contrast, patients with MYD88^MUT CXCR4^WT demonstrated deeper responses, shorter time to MR and improved PFS with ibrutinib therapy.³¹ More than 40 distinct CXCR4 variants have been identified in LPL or WM, with both the mutation type and burden of CXCR4 clone shaping the clinical phenotype and behaviour.^19,33 In C-terminal CXCR4 nonsense^(NS) versus CXCR4^WT, enhanced and prolonged AKT (protein kinase B) and BTK signalling follows activation of the CXCR4 receptor by SDF-1α.^22,34 This enhanced signalling is explained by the loss of essential regulatory regions in the distal region of the CXCR4 C-terminus, including key serine residues required for the attenuation of SDF-1α mediated CXCR4 signalling. This attenuation is typically mediated by β1 and β2 arrestin proteins, which assemble only when the serine residues are phosphorylated.^35,36 As demonstrated in CXCR4^S338* (c.1013C > G), this enhanced signalling leads to upregulation of genes associated with invasiveness, anti-apoptosis and drug resistance, thereby driving tumour proliferation, organ infiltration and drug resistance. Although translating into a shorter time to treatment, resistance to certain small molecule inhibitors, such as ibrutinib, OS remains unaffected by CXCR4^MUT.^31,34,37,38 Further, differences exist between MYD88^L265P with CXCR4^NS versus frameshift^(FS) C-terminal mutations, the former (when present in ≥ 25% of tumour cells)²⁵ associated with a higher bone marrow disease burden, elevated serum IgM levels, resulting in a higher incidence of symptomatic hyperviscosity requiring therapeutic intervention, and inferior response to ibrutinib therapy (worse MR, shorter PFS compared with CXCR4^WT).^19,20,25 The association of mutation status with extent of bone marrow infiltration was demonstrated in our cohort, where all cases with CXCR4 nonsense mutations demonstrated a diffuse infiltration pattern on histology. The reason for this difference in behaviour, based on mutation type, has been explained in cellular-level studies. Frameshift mutations were shown to alter the terminal amino acid sequence, resulting in a truncated CXCR4 protein with altered functional properties. In contrast, nonsense mutations triggered the messenger ribonucleic acid (mRNA) decay pathway, leading to markedly reduced levels of the truncated CXCR4 protein, thereby impairing CXCR4 receptor internalisation and signalling termination.³⁹

Risk stratification in WM is most commonly performed using the International Prognostic Scoring System for WM (IPSSWM), introduced by Morel et al.⁴⁰ In our South African public healthcare setting, however, the applicability is limited by inconsistent B2M testing. To address this, a simplified risk stratification model proposed by Zanwar and colleagues²⁶ was applied to better identify high-risk patients using more accessible clinical and laboratory parameters.²⁶ In our cohort, intermediate to high risk scores predominated among patients with MYD88 variants, with nearly half of those in the MYD88^MUT CXCR4^MUT subgroup classified as high risk. This group also exhibited other adverse features, including high IgM levels, diffuse bone marrow infiltration and the most pronounced cytopenias. Despite these high-risk features, no significant difference in OS was observed compared with other mutational subgroups, likely due to the cohort’s limited size.

Therapeutic intervention is not mandatory at the time of diagnosis of LPL or WM; the decision to initiate treatment remains a clinical judgement guided by disease-related symptoms.³ Recommended first-line therapeutic agents include rituximab (an anti-CD20 monoclonal human antibody), bendamustine (an alkylating chemotherapeutic agent) and BTKis. The class of drug and combination of agents depend, among other factors, on the specific disease-related complication; mutation status (MYD88, CXCR4, and TP53); and the availability of treatment options.³ Rituximab monotherapy yields lower response rates in LPL or WM compared to combination regimens, except when used to manage IgM-related complications. However, it should be avoided in the setting of elevated IgM levels or symptomatic hyperviscosity owing to the risk of rituximab-induced IgM flare.^3,41,42 For other disease-related complications, either chemo-immunotherapy (i.e., bendamustine–rituximab), continuous BTKi monotherapy (e.g., zanubrutinib or ibrutinib) or proteasome inhibitor-based regimens are utilised.^3,41,43,44

Regarding BTKi monotherapy, zanubrutinib has shown superior outcomes compared with ibrutinib in MYD88^NS CXCR4^MUT (improved MR and PFS) and TP53^MUT (improved MR, complete response, PFS).^31,45 For patients who develop resistance to first- and second-generation BTKi, newer alternatives, such as pirtobrutinib (a reversible BTKi) or venetoclax (a BCL-2 inhibitor), offer additional therapeutic options.⁴³ It is clear that routine evaluation of MYD88, CXCR4 mutation status, in patients requiring treatment is essential for guiding the selection of first-line therapy and to support efforts to expand access to BTKi and other targeted therapies in the South African public healthcare sector, where treatment options are largely restricted to rituximab (monotherapy or in combination with cyclophosphamide based regimens), with bendamustine availability inconsistent across the country.⁴³

Testing methods have advanced since the MYD88^L265P was first identified, with low-sensitivity Sanger sequencing (15%–20%)⁴⁶ surpassed by ASO-PCR (limit of detection 1%), albeit limited in detecting non-L265P variants.²⁷ The move is now towards droplet digital PCR, which shows deeper sensitivity (able to detect VAF as low as 0.1%), overcoming the limitation of low-infiltrated samples.^47,48 While next-generation sequencing offers the advantage of multigene analysis and the detection of mutations of prognostic importance in other genes, its diagnostic utility is limited by relatively low sensitivity (≥ 5%), leading to false negative MYD88 results in up to a third of bone marrow cases.^46,49 Importantly, the limitations of each testing method should be carefully considered when selecting a testing platform and interpreting the results.

In the South African public healthcare setting, CXCR4 variants are frequent in LPL or WM with high risk profiles most common among patients harbouring both MYD88 and CXCR4 variants.